UK schools are pressure-testing AI in classrooms. Here’s what their successes and failures reveal about using AI to boost productivity at work without losing the human core.
What UK Classrooms Get Right And Wrong About AI
Maths classes taught from 300 miles away. Deepfake "AI teachers" giving personalised feedback. Teachers striking because a screen replaced a colleague.
That’s not a dystopian pitch deck — it’s the UK education system in December 2025.
This matters for anyone thinking about AI, technology, work, and productivity because schools are pressure-testing the same questions every organisation faces: How far can you automate without breaking the human parts that actually make things work? Education just shows the tension more clearly.
In this post, I’ll walk through what’s happening in UK classrooms right now, why teachers are pushing back, and what this experiment tells us about using AI at work without losing trust, quality, or jobs.
The UK’s Real-Time Experiment With AI in Classrooms
The UK is running a live A/B test on AI in education. Some schools are going all-in on automation and remote teaching; others are drawing a hard line.
Here’s the snapshot:
- 85% of students and teachers globally now use AI tools, up from 66% in 2024.
- In the UK, 60% of educators already use AI in classrooms, mainly for personalised learning and grading.
- Yet only 12% of parents support AI in the classroom.
So usage is high, but trust is low. That gap is the real story.
Two case studies show the extremes:
- The Valley Leadership Academy, Lancashire – top-set maths groups are taught by a remote teacher 300 miles away, with a second adult in the room.
- Great Schools Trust (Liverpool, Warrington, Bolton) – using AI for marking, diagnosing learning gaps, and soon, deepfake “digital twins” of teachers to send individual feedback videos.
Supporters frame this as essential productivity: fix teacher shortages, reduce marking time, and give students more personalised support. Critics worry that “productivity” is just code for fewer human teachers, more screens, and weaker relationships.
Both are partly right.
Where AI Genuinely Boosts Teaching Productivity
The reality is simple: AI is brilliant at repetitive, pattern-based work and terrible at actual care. Education needs both.
High-ROI AI Uses in Schools
These are the workflows where AI and technology clearly help teachers work smarter, not just harder:
-
Marking and assessment
AI can handle multiple-choice and structured answers, flag likely misconceptions, and generate item-level analysis. Teachers get their evenings back instead of drowning in scripts. -
Identifying learning gaps
Systems can scan results across a class or year group and surface that “fractions with unlike denominators” or “simultaneous equations” are the problem, not algebra in general. That’s actionable data, fast. -
Content and lesson planning
Tools like AI lesson generators help teachers create differentiated tasks, examples, and retrieval practice in minutes instead of hours. The human still sets the intent and checks quality. -
Personalised practice for students
Adaptive platforms can serve questions at the right difficulty, at the right time, in the right format. That’s the kind of micro-tailoring one teacher can’t do manually for 150 students.
Used this way, AI in education does exactly what we want AI at work to do elsewhere:
Offload routine cognitive labour so humans can spend more time on judgment, creativity, and relationships.
This is the same pattern we see with AI for knowledge workers: draft generation, summarisation, analysis, and admin support. The schools that treat AI as a teaching assistant, not a replacement teacher, are getting the balance closer to right.
Where Schools Cross the Line: Remote Teachers and Deepfakes
If AI is so helpful, why are UK teachers striking?
Because there’s a difference between supporting the work and redefining the job.
Remote teaching 300 miles away
At The Valley Leadership Academy, top-set maths students have a remote teacher on screen and a separate adult in the room. On paper, that sounds workable: content expert on screen, classroom manager in person.
But teachers like Emily Cooke are asking the right question:
“Will your virtual teacher be there to dance with you at prom, hug your mum on results day, or high-five you in the corridor because they know you won the match last night?”
That’s not nostalgia. It’s workflow design. A teacher’s job isn’t just to transmit information:
- They notice who’s gone quiet.
- They spot the kid who’s pretending they understand.
- They tie today’s work to last week’s meltdown or last night’s football win.
Those micro-interactions are the relational glue that turns content delivery into real learning. When the “teacher” is a screen 300 miles away, that glue weakens.
Deepfake “AI teachers” and digital twins
Great Schools Trust is going further:
- AI marks work
- AI identifies gaps
- AI generates a deepfake “twin” of the teacher that records personalised feedback videos
On paper, this is hyper-efficient: every student gets targeted, detailed feedback in a format that feels 1:1.
Here’s the problem: the student’s experience is now mediated by a synthetic version of their teacher, not the actual human who knows them.
That raises three issues:
- Trust and authenticity – Students will eventually realise that some interactions are generated. When they do, what happens to their sense that “my teacher really sees me”?
- Likeness and consent – Who owns the model of a teacher’s face and voice? Can it be reused, repurposed, or kept after the teacher leaves?
- Job scope creep – Today it’s “opt-in digital twins for feedback”. Tomorrow it’s “the AI twin can handle revision lessons, parents’ evenings, induction videos… do we need as many staff?”
The tech itself isn’t the villain. The governance and incentives are.
The Real Fear: Productivity As a Cover for Austerity
Teachers aren’t anti-technology. Most already use AI in some form. What they don’t trust is the economic logic behind these decisions.
Schools are under pressure:
- National teacher shortages in key subjects
- Budget constraints year after year
- Political pressure to raise attainment without raising salaries
Introduce AI and remote teaching into that environment and the trajectory is predictable:
- Start as a short-term fix (“We can’t recruit a maths teacher; we’ll use a remote one”).
- Normalise it (“It’s just a small-scale initiative; it works fine”).
- Scale it (“If it works in top sets, why not middle sets? Why not other subjects?”).
- Restructure (“Do we really need that many on-site qualified teachers?”).
Parents pick up the risk instinctively. Only 12% support AI in the classroom. They’re not reacting to marking algorithms; they’re reacting to the idea that their child will end up in a hollowed-out school where productivity gains were captured as cost savings, not reinvested as better human support.
The lesson for any organisation adopting AI at work is clear:
If your AI strategy is primarily about reducing headcount, your people will resist – loudly and rationally.
If it’s about reducing drudgery and increasing quality, and you can prove that in practice, adoption gets much easier.
A Better Framework: Augment First, Automate Last
There’s a healthier way to bring AI into education (and into any workplace): augment first, automate last.
Step 1: Automate the invisible grunt work
Start with the tasks nobody values but everyone hates:
- Marking routine assessments
- Creating differentiated worksheets or quizzes
- Drafting reports, feedback comments, or communications
- Summarising student data and spotting patterns
Impact on students: better feedback, faster responses, more prepared lessons.
Impact on teachers: more time for planning, mentoring, and real conversations.
Step 2: Use AI to enhance human interaction, not replace it
Some smart uses here:
- AI-generated conversation prompts for 1:1 mentoring or coaching
- Behaviour and wellbeing alerts based on patterns in attendance, performance, and engagement, so teachers can intervene earlier
- Real-time translation or support for EAL (English as an additional language) students so they can participate more fully in class
The teacher is still the protagonist. AI just feeds them better information.
Step 3: Draw bright red ethical lines
This is where many AI projects fail: they treat ethics as a footnote, not a design constraint.
For schools (and by extension, companies), those lines might look like:
- No use of AI deepfakes of staff without explicit, revocable consent
- No AI-only “teachers” in place of qualified on-site staff for core subjects
- Clear policies on data usage, model training, and retention
- Transparency to students and parents about where AI is used and what it does
Once those lines are public and enforced, trust goes up. People are much more willing to accept automation inside a clearly safe box.
What This Means For Anyone Using AI at Work
The UK classroom debate isn’t just a schools story; it’s a mirror for every team adopting AI.
Here’s what I’d pull out if I were leading AI adoption in any organisation:
-
Start where AI is obviously helpful.
Go after the “marking and admin” equivalents in your world: report generation, inbox triage, data analysis, drafting, scheduling. -
Protect the human moments on purpose.
In teaching, that’s corridor conversations and results day hugs. At work, it’s 1:1s, performance feedback, difficult conversations, creative workshops. AI shouldn’t touch those. -
Be explicit that AI is for capacity, not cuts.
If productivity gains go directly into layoffs, staff will resist everything you try next. If they go into better service, shorter hours, or new projects, people lean in. -
Co-design with the people doing the work.
The teachers striking in Lancashire aren’t being precious; they’re telling leaders where the solution design broke the job. The same is true for any profession. -
Measure the right outcomes.
In schools, that’s not just grades; it’s engagement, attendance, wellbeing, and teacher retention. In business, it’s not just throughput; it’s quality, trust, and staff satisfaction.
I’ve found that teams who treat AI as a “junior colleague” they’re training – rather than a mysterious oracle they’re forced to obey – get better results and far less pushback.
Where Education Goes Next – And Why You Should Care
UK schools are a preview of how AI and technology will reshape work and productivity across every sector:
- The tools will keep improving: more capable models, better interfaces, more automation.
- The pressure will keep increasing: budgets, shortages, expectations.
- The real differentiator will be how leaders choose to use AI.
If education can find a balance where AI handles the repetitive cognitive work and humans double down on empathy, judgment, and relationships, that’s a template worth copying.
If instead we end up with remote screens, synthetic faces, and fewer adults in the room, we’ll have gained some productivity and lost something we can’t easily rebuild.
The question for schools, and for every workplace, is the same:
Are you using AI to make the human work stronger, or to quietly make it smaller?
The organisations that get this right won’t just work faster — they’ll work smarter in a way people actually want to be part of.